Distributed Online Optimization With Dynamic Coupling Constraints Under Time-Varying Communication Delays

Abstract

This paper considers distributed online optimization with dynamic inequality constraints under time-varying communication delays. A group of agents cooperatively estimates an optimal strategy by exchanging the information on the loss value that is sequentially disclosed to each agent. We develop a distributed primal-dual algorithm for an enlarged multiagent network with delayed agents that handle the delayed information. To analyze the optimality and feasibility of the algorithm, a dynamic regret function and an accumulated fit function are considered. We show that both the dynamic regret and fit functions achieve sublinear bounds even in the presence of communication delays. The numerical example verifies the theoretical claims of the sublinearity of the proposed method.

Keywords

• Distributed optimization
• multiagent system
• directed network
• communication delay